Specifically, in various situations, the recipient needs to know that
followups go to T, and replies go to R, where R isn't contained in T.
No, the recipient needs to _decide_ where his particular reply goes,
based on the content of his reply -- not take the sender's blind
recommendation on faith.
This recipient would be perfectly happy to take the sender's
recommendation on faith (with the aid of a single button that was known to
produce that effect).
I have no problem with your being able to make that choice. I have a big
problem with user agents or an architecture that _expects_ users to take
a sender's recommendation on faith.
Keith